🟢️ Stable releases of the data and code found in this repository are available on Zenodo (https://doi.org/10.5281/zenodo.17536656). 🟢️

🔗 The output from this repo can be viewed at https://digital-wellbeing.github.io/open-play/. 🔗

This repo hosts the data and descriptive analysis code for our project on video game play and wellbeing.

The best starting point for understanding the data presented here is to look at codebook.xlsx, which describes the variables in both the cleaned and raw dataset.

Other key elements of the repo are:

• codebook.xlsx, which describes the variables in both the cleaned and raw dataset
• data-process.qmd, which processes the raw data into cleaned data
• index.qmd, which generates manuscript documenting the dataset
• data/raw, which contains the minimally processed, pseudonymized raw data
• data/clean, which contains cleaned data ready for analysis

To reproduce the primary data manuscript, run quarto render. To reproduce the data manuscript and the simulations, run quarto render --profile sim.

Data files are saved as .csv.gz for size. These can either be unzipped and opened in a spreadsheet program, or read directly into R using readr::read_csv() or Python using pandas.read_csv().

Contributors should ensure they are up-to-date on the main branch, and that they have installed all the necessary packages from renv by running renv::restore() in R.

Contributors should then:

• create a new branch for their work
• make any desired changes (for readability, we recommend making frequent small commits focused on changing an individual feature, but these can be pushed less frequently)
• run quarto render locally to ensure their changes do not break the build
• submit a pull request when they are ready to merge their changes back into main.

Rendering locally will update the _freeze folder, which in turn will update the GitHub Pages site when the PR is merged; make sure PRs commit changes to _freeze or they will fail the build check.

In R/, we have a number of helper functions used in the analysis scripts, as well as unit_tests (currently implemented for the key function used to clean Xbox data, canonicalize_xbox_sessions(). Supplementary features can be added by adding new scripts to this folder.

For the Stage 1 Registered Report associated with this project, we did extensive simulations to validate our data collection and analyses. These remain present in the current repo, and can be run by using the --profile sim flag when rendering the quarto documents. This will then generate a book instead of a website.

Data for the simulations is found in data/synthetic.

The simulation pipeline consists of:

• [not run] 0_generateSyntheticData.qmd, documents the internal process we used to generate synthetic data
• 1_preprocess.qmd cleans the data and calculates relevant derived variables (e.g., mean scores, play behavior metrics, and so on).
• 2_basicNeeds.qmd presents the analysis code for Study 1: the relationship between basic needs and video game play.
• 3_sleep.qmd presents the analysis code for Study 2: the relationship between sleep and video game play.
• 4_genres.qmd presents the analysis code for Study 3: the relationship between video game genres and video game play.